Mismatch

FIGURE 16.1. Drift Toward Affluence

Sources: Authors’ calculations of data from National Longitudinal Study (1972); High School & Beyond (1982); National Longitudinal Education Study (1992).

When we talk about class in America, we are talking about a somewhat fuzzy concept; Americans obviously do not sort themselves into clearly distinct social strata based on accent, clothing, or habits. In this chapter we treat social status as a continuum ranging from the poor to the rich, from people with an eighth-grade education to others with doctorates or professional degrees. Social scientists usually measure “socioeconomic status” (SES) with some combination of information about a person’s education, her occupation, and her household income. When they analyze the SES of college students, they generally measure the education, occupation, and/or income of each student’s parents. An SES index puts these measures into a scale so that we can talk about someone whose parents rank in, say, the bottom quartile of SES or the top tenth. We know these rankings mean something because one can predict with reasonable accuracy many aspects of a person’s life by knowing their SES or (before they are independent) the SES of their parents. People tend to marry others with similar SES, live in neighborhoods where their SES predominates, and end up with jobs predicted by their parents’ SES.

Measures of the SES of students attending elite colleges and professional schools show that their backgrounds are dramatically skewed toward affluence. As Figure 16.2 shows, roughly three-quarters of students at elite colleges are from families in the top quarter of SES, whereas only 3 percent come from the bottom quarter and under 10 percent come from the bottom half. Put differently, this means that a boy or girl who grows up in a top-quartile family is about twenty-five times more likely to end up attending an elite college than someone from the bottom quartile. The class background of students is even more skewed at elite law schools.

FIGURE 16.2. Not-So-Diverse

Source: Analysis of elite colleges is from Carnevale & Rose, “Socioeconomic Status, Race/Ethnicity, and Selective College Admissions,” (2004); analysis of law schools is from Sander, “Class and American Legal Education,” (2011).

We can intuitively sense from this data that low-SES students are much more underrepresented in the ranks of selective colleges than are, for instance, blacks or Hispanics. A random black high school senior is perhaps 30 percent less likely than a random white high school senior to end up at Harvard, but a senior from the bottom quartile of SES is dozens of times less likely to end up at Harvard than a senior from the top quartile or top tenth of SES.

Indeed, even if we consider college education more broadly, the disparity in opportunities is striking. Using the National Educational Longitudinal Study (NELS), the federal government’s major effort in the 1990s to study such things as the transition from high school to college, we can compare the chances of different groups of high school graduates advancing to a four-year college. When we control for academic preparation, black high school graduates are about 30 percent more likely than comparable whites to attend a four-year college. But persons of all races from the bottom 20 percent of the SES distribution are about 70 percent less likely than otherwise comparable people from the top 20 percent of the SES distribution to attend a four-year college. Ironically, the same data shows that blacks are less likely than comparable whites to actually receive a bachelor’s degree—a plausible consequence of mismatch. In other words, racial preference policies (and their cascading effects through the system) have more than leveled the playing field for blacks in college access. However, our system does a poor job on two other counts: providing access to able students from the lower SES ranges and helping blacks succeed after they get to college.

Especially in the context of our book’s other findings, these patterns from the NELS are unmistakable signs of a racial preference system gone badly awry. And the picture becomes all the more distressing when we look closely at who actually receives preferences. As we noted at the beginning of the chapter, 92 percent of the blacks attending elite colleges (in the NELS sample) came from the families in the top half of the American SES distribution (which means the top third of the black SES distribution). The NELS sample is small, but data from the “After the JD” (AJD), a database we discussed in Chapters Four and Five, has detailed SES information on a much larger sample of law students who finished law school in 1999 or 2000. Analysis of this data shows that 89 percent of blacks entering elite (top twenty) law schools—a large majority of whom received large racial preferences—came from families in the top half of the American SES distribution. Sixty-six percent came from families in the top quarter, and 43 percent came from families in the very elite and affluent families in the top tenth. Hispanic law students were more likely to come from working-class backgrounds (and received smaller preferences), but nonetheless, half of the Hispanic students at elite law schools came from families in the top quarter of the SES distribution.

When we look more closely at students receiving preferences, their “representativeness” dissolves even further. Ten years ago black students at Harvard Law School organized a sort of self-study. They found that only 30 percent of black students at the school had four African American grandparents. Another third of the blacks were from interracial families, and the rest were either foreign born or the children of foreign-born people of West Indian and/or African descent. More systematic studies have found that nearly 40 percent of black Ivy League undergraduates are first- or second-generation immigrants. And for mixed-race students (recall Natasha Scott from Chapter Twelve), the potential for inflated numbers has become so significant that the federal government changed its rules last year, now requiring colleges to report mixed-race students separately. “In the past,” reports Jeff Brenzel, dean of undergraduate admissions at Yale, “if a student self-identified as both African-American and white, or both African-American and Hispanic, the student would be counted as African-American.”

The temptation for admissions officers to admit, in greatly disproportionate numbers, affluent blacks, mixed-race blacks, and foreign-born blacks arises from three simple facts. First, the academic indices of each of these groups are much higher, on average, than those of American blacks in general. Second, they are easy to find; indeed, the admissions offices don’t need to do any outreach at all—students in these groups will tend to seek out and apply to elite schools. Third, if one is focused on a numbers game, then anybody who can be given the appropriate racial identification will serve. (One of us once witnessed an admissions office, particularly desperate to get up its “black” numbers, count someone of Egyptian descent as African American.)

As America becomes more multiethnic and intermarriage grows in popularity, our multiracial population has soared, growing by 36 percent from 2000 to 2009 and projected by the census to grow another 21 percent by 2015, to over 6.4 million. Black immigration continues to grow as well; first-and second-generation black Americans (who tend to be highly educated) now make up close to 20 percent of the black population. Under these demographic circumstances, even admissions officers who want to be racially inclusive face increasingly difficult and arbitrary distinctions. Who should count as black? And these difficulties are even greater when we consider Hispanics, Southeast Asians, or American Indians, all of whom have still higher cross-racial marriage rates.

There are, as well, the increasingly important intersections between class and race. This flows in two directions. On the one hand, middle- and upper-middle-class blacks have increasingly entered the mainstream of American life. They are likely to be college educated, have professional or managerial jobs, live in attractive neighborhoods, and have cross-racial social networks. When we control for measures of learning and academic preparation, blacks with college degrees have the same or higher earnings and are more likely to have professional jobs than are comparable whites. On many matters they have attitudes about a host of social and economic issues similar to those of their middle- and upper-middle-class white neighbors and work colleagues, especially if they live on the East or West coast.

But “intersectionality” occurs in a completely different way for those in the bottom half of the American income distribution. Low-income blacks tend to be substantially worse off than whites with similar incomes. Low- and moderate-income blacks are very likely to live in highly segregated neighborhoods. A very high proportion of their neighbors are likely to be poor. The local school is likely to be terrible. Crime rates are higher; two-parent families are a rarity. As Eugene Robinson powerfully argues in his 2011 book Disintegration , social policy needs to (but often fails to) distinguish between America’s distinct strata of blacks; for all of the enthusiasm about diversity, lower-and working-class blacks are the ones who need help but are often utterly overlooked.

As matters stand, our racial preference system is poorly targeted. Whether the goal is to achieve a meaningful representation of America’s population, to create classrooms that have a diversity of life experiences, or to improve social mobility, we are not doing a passable job.

Defenders of the status quo who concede these huge disparities often suggest that greater consideration of class might be desirable, but is unworkable because there are simply not enough low-SES students who can cut the mustard.

Not true. As we noted earlier, high school seniors in the bottom fifth of the SES distribution are 70 percent less likely to attend a four-year college than academically similar seniors in the top fifth of the SES distribution. That means that when we are comparing two seniors—one low-income and one high-income—who have the same measureable level of academic preparation, the high-income senior is three times as likely to go to a four-year college.

An even more compelling way of seeing this point is captured in Figure 16.1. Using the College Board data described in Chapter Eight, we examined the pool of all students taking the SAT who, under the pre-2006 scoring system, had a combined score of 1200 or more on the math and verbal portions of the test. A 1200 score puts a student into the top 8 or 9 percent of all high school students—well within the pool of students that even the most elite schools consider (under preference programs). This pool is sorted by race and quintiles of socioeconomic status; the number in each cell shows the proportion of those students who applied to any of a loosely defined “top ten” undergraduate colleges (the Ivy League plus Duke and Stanford). These data show that a high-scoring black student in the bottom SES quintile had only a 4 percent chance of applying to any of these schools, whereas a high-scoring black student in the top SES quintile had a 48 percent chance. The disparities are smaller but still striking for white and Hispanic students. Only among Asians, whose parents often have a fierce determination to maximize the educational achievement of their children, are the disparities seriously blunted.

FIGURE 16.3. Untapped Potential

Source: Authors’ calculations of data obtained from the College Board. Top-tier colleges include the Ivy League, Duke, and Stanford.

Figure 16.3 speaks volumes about the landscape of opportunity for promising high school students and about the poor job that college admissions officers are doing in finding “diamonds in the rough” outside prep schools and high schools attended by the upper-middle class. In fact, this analysis understates our point in an important way. An upper-middle-class student who does reasonably well in high school is almost on a default path to college and, thus, automatically in the pool of students who take the SAT or ACT along that path. A student growing up in a poor household is on the default path not to attend college—or if college is considered at all, to attend the local community college. Therefore, a substantial number of talented low-SES students never take the SAT and, thus, never make it into the pool this data captures.

Moreover, as we saw in Part III, schools such as the University of California that have been constrained to avoid or minimize racial references end up doing a dramatically better job than their peer institutions in assembling minority cohorts that are far more representative of the populations they purport to represent. Our preference systems are simply missing a very large pool of talented, low-SES students.

In the wake of Prop 209 one of us (Sander) helped to persuade his law school (UCLA) to adopt a system of race-neutral, class-based preferences. The experiment was controversial within the school, and it was modified after its first year. But by the criteria discussed in this chapter (and to many of its opponents in retrospect), the experiment was a stunning success. With an average preference of about one-third the size of the racial preferences used before Prop 209, the law school assembled a class that actually resembled the socioeconomic distribution in America. More than a third of the students came from the bottom half of the SES distribution; more than a quarter of the students were the first in their family to attend college. The class was nearly a third nonwhite and included more underrepresented minorities than any other race-neutral method that had been adopted at any UC professional school. And, very importantly, the students were academically successful. The cohort admitted through this program (the graduating class of 2000) achieved the highest bar passage rate in the school’s history.

All of this was achieved without a significant outreach effort or a very broad awareness among low-SES college students of what we were doing; with such efforts in place, we plausibly could have admitted an even stronger class. (UCLA still uses significant class-based preferences, though in a somewhat retrenched form, and has managed to advance steadily vis-à-vis its law school peers in the average academic index of its students.)

A hallmark of the UCLA experiment was its use of measures that captured the intersections between race and class that we mentioned earlier. Conventional measures of SES—which, as we have noted, usually are limited to measures of income, education, and occupation—understate levels of disadvantage for Hispanics and blacks. A black household in the middle of the income distribution has, on average, only a fraction of the assets held by a typical white household with the same income. Segregation concentrates low-SES Hispanics and blacks in poorer neighborhoods with weaker institutions.

To address those intersections, the law school used eight measures in constructing a measure of socioeconomic status: In addition to the education of each parent and the parent’s income, it also measured the parent’s assets and several indices of social conditions in the zip code in which the applicant lived during high school, such as the poverty rate and high school dropout rate.

Including such factors meant that blacks and Hispanics with a given level of academic index were much more likely to be admitted than were whites and Asians. Considering only the broadly defined SES of applicants and not their race, UCLA Law School admitted 21 percent of its black applicants and 24 percent of its Hispanic applicants with good but not outstanding academic indices, compared with 7 percent of its white applicants and 7 percent of its Asian applicants in the same academic index range. The school achieved substantial racial diversity, a unique level of socioeconomic diversity, and a high level of academic success for its students.

As we discussed in Part III, the UC system as a whole also experimented with class-based preferences after Prop 209 and had similar successes. At the end of the 1990s the UC system had managed to achieve systemwide black and Hispanic enrollment levels that were close to their pre-209 levels (with many minority students cascaded to less elite campuses), had improved black and Hispanic academic achievement, and had achieved far higher levels of socioeconomic diversity than had comparable schools elsewhere in the nation. More than 30 percent of Berkeley and UCLA undergraduates qualified for Pell Grants (a form of federal educational assistance generally available to students from roughly the bottom half of the income distribution); at other elite colleges around the country the Pell Grant rate hovered around 10 percent. The problem with the UC experiment came after 2000, as the professional schools and eventually the colleges were tempted to introduce more and more aggressive proxies for race itself. Those steps tended to reintroduce or exacerbate the mismatch problem.

This leaves us with a vital question: Can we increase SES diversity in American higher education without simply exacerbating the mismatch problem? We think the answer is yes, for two reasons.

First, there is a great deal that universities can do to increase SES diversity without any use of preferences. They can do a much better job of finding the talented students excelling at nonelite high schools that are off the radar from outreach officers. They can do a better job of simply recognizing that the applications of students with socioeconomic disadvantage may require a closer, different kind of scrutiny than that applied to affluent applicants. Many admissions officers focus on how “interesting” a student is—how original her essay might be, how many countries she has visited, whether she has done innovative charity work or an unpaid internship with, say, an environmental group. Giving weight to these sorts of experiences may help produce an eclectic entering class, but it is important to keep in mind that low-SES students are far less likely to offer them: They are more likely to work at uninteresting jobs during the summer (to help support their families), and they are far less likely to have some kind of résumé coach.

It is well documented that expensive private colleges have been particularly guilty parties in the pervasive grade inflation that has swept over higher education in the past forty years. In the applications to graduate and professional schools, students who have attended private colleges are more likely to have inflated grades than otherwise similar students who have attended public universities of the same academic quality. If admissions officers do not take this distinction into account—and they generally do not—they will disfavor the low-SES applicants who disproportionately attend public universities. It is plausible that similar mechanisms reward affluent college applicants who have attended undistinguished but expensive private schools.

Higher education institutions can also do much more to make themselves affordable to low-SES applicants. For most of the past forty years college tuitions have risen much faster than the rate of inflation. Some very elite colleges (e.g., Harvard and Princeton) have not only practiced need-blind admissions for a long time but have in the past decade introduced dramatic policies that promise tuition-free education to low-, moderate-, and even many middle-income families. If more schools move in this direction, it would substantially increase enrollments of low-SES students without the use of preferences.

If we again consider law schools as an example, recent data show two disturbing patterns. First, among all students enrolled at a broad cross-section of law schools in the mid-1990s, the white students with the lowest academic indices (compared to others at the same school) were those with the highest SES. The implication is that the law school admissions systems tended, on balance, to favor affluent applicants over others and, thus, admitted more affluent white students with slightly lower academic qualifications over other whites. (These patterns existed alongside much larger racial preferences aimed at blacks and Hispanics.)

Second, high-SES blacks at law schools nationwide receive, on average, four times as much scholarship money as do low-SES whites. This suggests that law schools deploy disproportionate resources toward bidding wars for black students while providing very little need-based aid. (We saw in Chapter Fourteen how George Mason Law School engaged in this behavior while under relentless pressure from the ABA.) Shifting available aid to focus more on need and less on diversity gamesmanship would increase SES diversity.

There is, in short, a great deal that colleges and universities can do simply to make the playing field more level for low-SES candidates. The question of whether they should go further and actually grant preferences based on SES is a harder one that we return to in Chapter Eighteen. We know from the experiments described above that sharp increases in socioeconomic diversity can be achieved with preferences much smaller than those typically used to pursue racial diversity. We also know that mismatch is fundamentally related not to whether preferences are tied to race, SES, or athletic ability but rather to the size of the preference. What we do not know is how large a preference is too large. Small preferences may well be harmless or may even foster positive peer effects. These uncertainties create difficult policy tradeoffs—tradeoffs we will try to navigate in Chapter Eighteen.